I gain about .4 or 4% more performance per 10% increase on clock speeds, I tried that out
Printable View
I gain about .4 or 4% more performance per 10% increase on clock speeds, I tried that out
thunderbird, what if i told you i get 12,700 marks with all of my hardware at stock speeds except for my geforce 4? my 2100+ at 1.73, (133x13.0) but with my video card overclocked? Wouldn't that prove that you need a combination of fsb, clock speed, and a fast video card to score high? So really, fsb is not everything. My old tbird was holding me back because it couldn't do higher clock speed. Even though it did 200 fsb.
wow not bad. of course if you ran 200x8 for 1.6GHz youd blow a stock xp2100 outta the water big time :) also use that kt333 of yours and run 166fsb for 2.16GHz, that should get you like 14k marks which is plenty :)
I could swear I had more IQ points before reading this...Quote:
Originally posted by Thunderbird1GHz
raw MHz matters in specific apps, obviously superPI and whatever does not use the video card or the ram, just the cpu for number crunching. Games use everything and games also love fast ram and lots and lots of ram bandwith this is why an increase of 1MHz to the fsb gains me more than 10MHz increase on the cpu. 33fsb is equivelent to 333MHz, thus xp3000s run 200fsb for 2.2GHz while xp2700 at 166fsb and 2.16GHz. 200fsb and 1800MHz is just as fast in games as 166fsb and 2.16GHz or stock xp2700 speeds :)
ummmm...last i checked, .4=40%Quote:
I gain about .4 or 4% more performance per 10% increase on clock speeds, I tried that out
Dude, there's no way a T-Bird at 200x8 could smoke a 2100+ at 166x13
I don't get you, people toss allllll this info at you and actually back it up with proof and somehow you find a way to make yourself think the bs you post is true.
Though I do have to commend you. Through your misguided attempts of showing off your rig, I have learned some valuable stuff from the others posting on your topics.
err I said stock xp2100. 1.6GHz on 200fsb has a 667MHz advantage over 133fsb or 333MHz over 166 so 200x8=1600 or 166x11.5=1909(a bit slower) or 133x17=2261. 1GHz at 200fsb is like a stock xp2000 at 1.66GHz on 133fsb :)
Quote:
err I said stock xp2100. 1.6GHz on 200fsb has a 667MHz advantage over 133fsb or 333MHz over 166 so 200x8=1600 or 166x11.5=1909(a bit slower) or 133x17=2261. 1GHz at 200fsb is like a stock xp2000 at 1.66GHz on 133fsb
>>>>>>>>>>>>>>>>>>>
Whaaaa????????????????????
You need another hobby.
baldy:cool:
Quote:
Originally posted by bldegle2
Whaaaa????????????????????
Ditto. Only he understands the crap that he spreads around the forum.
What workload? If you are talking about 3Dmark again, you can stop right there. It isn’t a good processors benchmark at all, and doesn’t reflect the way most applications will respond to changes in clock rate.Quote:
Originally posted by Thunderbird1GHz
I gain about .4 or 4% more performance per 10% increase on clock speeds, I tried that out
Quote:
Originally posted by Moridin
What workload? If you are talking about 3Dmark again, you can stop right there. It isn’t a good processors benchmark at all, and doesn’t reflect the way most applications will respond to changes in clock rate.
Now you see why we all get pissed and flame him? He pulls numbers out of his a.s.s all the time, he won't stop comparing EVERYTHING to 3Dmark, he would refer to it every 2 minutes and base his all life on his 3DMark score :)
I also used quake3 in 640x480 which should be a good indicator of cpu performance scaling. I dont care if it runs prime95 faster, I just play games :p and I tried several games, it scales so poorly a faster cpu isnt really justified. I have calculated Ill need 210fsb and 2.42GHz and rare as hen's tooth ram thatll run 210fsb at 2-2-2-6-t1 and high end air cooling all for the possibily of breaking 14k. I would be better and cheaper off getting a $200 radeon9700 non pro instead and that could get me 15k easy or more but no need for neither
Prime 95 is not a benchmark, you don't do it "fast". You have to pass it, that's all, pass it, and pass it without it crashing.
So call him on it. Tell him why his reasoning is wrong, and what logic mistakes he is making. This is perfectly acceptable, and far more productive then flaming.Quote:
Originally posted by Ignitionator
Now you see why we all get pissed and flame him? He pulls numbers out of his a.s.s all the time, he won't stop comparing EVERYTHING to 3Dmark, he would refer to it every 2 minutes and base his all life on his 3DMark score :)
So call him on it. Tell him why his reasoning is wrong, and what logic mistakes he is making. This is perfectly acceptable, and far more productive then flaming. [/B][/QUOTE]
Uhhhhhh, hasn't that been done more times than one can count (the
polite and logical illustration of errors in judgement)???
It's an ok indicator of CPU performance, and a sub par indicator of CPU scaling. It is CPU dependant, but it's not exactly typical of newer applications. When it was a current game it scaled at roughly 0.7 today it scales at around 0.5, but the ridiculously high frame rate means it does become more memory bound then a more modern application.Quote:
Originally posted by Thunderbird1GHz
I also used quake3 in 640x480 which should be a good indicator of cpu performance scaling.
http://www.tech-report.com/reviews/2...2/index.x?pg=8
If you look in the following article, a P4C 2.4 GHz - 3.2 GHz For example 33% clock rate increase = 17% performance increase in quake. If you up the FSB of the 2.8 GHz 50% from 533MHz to 800 MHz you only get a 9% increase. This gives it a scaling factor of less then 0.2 for FSB speed.
Remember, all applications tend to become more memory bound over time because as the CPU becomes faster it crunches through the data more quickly. This inevitably reverts to a more normal balance in the next generation of software because the faster processor allowed developers to use more processor intensive algorithms, which in turn shifts the load back to the processor.
Despite this, Quake is still more processor bound then memory bound. With more complex applications, particularly ones with more complex physics, you will get back up to the 0.6 to 0.7 scaling I referred to earlier. Again, if you are GPU bound, processor clock rate will not matter at all, though memory/AGP bandwidth still could if you don't have enough graphics memory.
You can't force people to learn, that doesn’t give you the right to be disrespectful to them. If you are afraid someone will take bad advice based on his posts then by all means point out the flaws. He may not change his position, but if your response is well though out the logic errors will be apparent, even to the newbies.Quote:
Originally posted by hwmicre
Uhhhhhh, hasn't that been done more times than one can count (the
polite and logical illustration of errors in judgement)???
Other then that, you may simply not want to respond at all if he isn't going to listen. Remember, you can say your piece without forcing someone to agree with you. If you or someone else responds thoughtfully, and he fails to bring up any new points then why respond at all?
You do understand that Ram that can do 2-5-2-2 at 210fsb is not rare at all right?Quote:
Originally posted by Thunderbird1GHz
I have calculated Ill need 210fsb and 2.42GHz and rare as hen's tooth ram thatll run 210fsb at 2-2-2-6-t1
Either Corsair PC-3500 or HyperX or many many other types of Ram will do this. And price has come down significanly. You can get Corsair PC-3500 512 meg now for like $140.
Unfortunately for me, I bought a gig a the stuff for $480 right before the prices started falling.
140 a stick is still a bit much. I'm waiting for the stuff to hit 100, hopefuly it will :-/